dom/media/AudioChannelFormat.h

   1 /* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
   2 /* vim:set ts=2 sw=2 sts=2 et cindent: */
   3 /* This Source Code Form is subject to the terms of the Mozilla Public
   4  * License, v. 2.0. If a copy of the MPL was not distributed with this
   5  * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
   6 #ifndef MOZILLA_AUDIOCHANNELFORMAT_H_
   7 #define MOZILLA_AUDIOCHANNELFORMAT_H_
   8
   9 #include <stdint.h>
  10
  11 #include "mozilla/PodOperations.h"
  12 #include "nsTArrayForwardDeclare.h"
  13 #include "AudioSampleFormat.h"
  14 #include "nsTArray.h"
  15
  16 namespace mozilla {
  17
  18 /*
  19  * This file provides utilities for upmixing and downmixing channels.
  20  *
  21  * The channel layouts, upmixing and downmixing are consistent with the
  22  * Web Audio spec.
  23  *
  24  * Channel layouts for up to 6 channels:
  25  *   mono   { M }
  26  *   stereo { L, R }
  27  *          { L, R, C }
  28  *   quad   { L, R, SL, SR }
  29  *          { L, R, C, SL, SR }
  30  *   5.1    { L, R, C, LFE, SL, SR }
  31  *
  32  * Only 1, 2, 4 and 6 are currently defined in Web Audio.
  33  */
  34
  35 enum {
  36   SURROUND_L,
  37   SURROUND_R,
  38   SURROUND_C,
  39   SURROUND_LFE,
  40   SURROUND_SL,
  41   SURROUND_SR
  42 };
  43
  44 const uint32_t CUSTOM_CHANNEL_LAYOUTS = 6;
  45
  46 // This is defined by some Windows SDK header.
  47 #undef IGNORE
  48
  49 const int IGNORE = CUSTOM_CHANNEL_LAYOUTS;
  50 const float IGNORE_F = 0.0f;
  51
  52 const int gMixingMatrixIndexByChannels[CUSTOM_CHANNEL_LAYOUTS - 1] = {0, 5, 9,
  53                                                                       12, 14};
  54
  55 /**
  56  * Return a channel count whose channel layout includes all the channels from
  57  * aChannels1 and aChannels2.
  58  */
  59 uint32_t GetAudioChannelsSuperset(uint32_t aChannels1, uint32_t aChannels2);
  60
  61 /**
  62  * DownMixMatrix represents a conversion matrix efficiently by exploiting the
  63  * fact that each input channel contributes to at most one output channel,
  64  * except possibly for the C input channel in layouts that have one. Also,
  65  * every input channel is multiplied by the same coefficient for every output
  66  * channel it contributes to.
  67  */
  68 const float SQRT_ONE_HALF = 0.7071067811865476f;
  69
  70 struct DownMixMatrix {
  71   // Every input channel c is copied to output channel mInputDestination[c]
  72   // after multiplying by mInputCoefficient[c].
  73   uint8_t mInputDestination[CUSTOM_CHANNEL_LAYOUTS];
  74   // If not IGNORE, then the C channel is copied to this output channel after
  75   // multiplying by its coefficient.
  76   uint8_t mCExtraDestination;
  77   float mInputCoefficient[CUSTOM_CHANNEL_LAYOUTS];
  78 };
  79
  80 static const DownMixMatrix gDownMixMatrices[CUSTOM_CHANNEL_LAYOUTS *
  81                                             (CUSTOM_CHANNEL_LAYOUTS - 1) /
  82                                             2] = {
  83     // Downmixes to mono
  84     {{0, 0}, IGNORE, {0.5f, 0.5f}},
  85     {{0, IGNORE, IGNORE}, IGNORE, {1.0f, IGNORE_F, IGNORE_F}},
  86     {{0, 0, 0, 0}, IGNORE, {0.25f, 0.25f, 0.25f, 0.25f}},
  87     {{0, IGNORE, IGNORE, IGNORE, IGNORE},
  88      IGNORE,
  89      {1.0f, IGNORE_F, IGNORE_F, IGNORE_F, IGNORE_F}},
  90     {{0, 0, 0, IGNORE, 0, 0},
  91      IGNORE,
  92      {SQRT_ONE_HALF, SQRT_ONE_HALF, 1.0f, IGNORE_F, 0.5f, 0.5f}},
  93     // Downmixes to stereo
  94     {{0, 1, IGNORE}, IGNORE, {1.0f, 1.0f, IGNORE_F}},
  95     {{0, 1, 0, 1}, IGNORE, {0.5f, 0.5f, 0.5f, 0.5f}},
  96     {{0, 1, IGNORE, IGNORE, IGNORE},
  97      IGNORE,
  98      {1.0f, 1.0f, IGNORE_F, IGNORE_F, IGNORE_F}},
  99     {{0, 1, 0, IGNORE, 0, 1},
 100      1,
 101      {1.0f, 1.0f, SQRT_ONE_HALF, IGNORE_F, SQRT_ONE_HALF, SQRT_ONE_HALF}},
 102     // Downmixes to 3-channel
 103     {{0, 1, 2, IGNORE}, IGNORE, {1.0f, 1.0f, 1.0f, IGNORE_F}},
 104     {{0, 1, 2, IGNORE, IGNORE}, IGNORE, {1.0f, 1.0f, 1.0f, IGNORE_F, IGNORE_F}},
 105     {{0, 1, 2, IGNORE, IGNORE, IGNORE},
 106      IGNORE,
 107      {1.0f, 1.0f, 1.0f, IGNORE_F, IGNORE_F, IGNORE_F}},
 108     // Downmixes to quad
 109     {{0, 1, 2, 3, IGNORE}, IGNORE, {1.0f, 1.0f, 1.0f, 1.0f, IGNORE_F}},
 110     {{0, 1, 0, IGNORE, 2, 3},
 111      1,
 112      {1.0f, 1.0f, SQRT_ONE_HALF, IGNORE_F, 1.0f, 1.0f}},
 113     // Downmixes to 5-channel
 114     {{0, 1, 2, 3, 4, IGNORE},
 115      IGNORE,
 116      {1.0f, 1.0f, 1.0f, 1.0f, 1.0f, IGNORE_F}}};
 117
 118 /**
 119  * Given an array of input channels, downmix to aOutputChannelCount, and copy
 120  * the results to the channel buffers in aOutputChannels.  Don't call this with
 121  * input count <= output count.
 122  */
 123 template <typename SrcT, typename DstT>
 124 void AudioChannelsDownMix(Span<const SrcT* const> aInputChannels,
 125                           Span<DstT* const> aOutputChannels,
 126                           uint32_t aDuration) {
 127   uint32_t inputChannelCount = aInputChannels.Length();
 128   uint32_t outputChannelCount = aOutputChannels.Length();
 129   NS_ASSERTION(inputChannelCount > outputChannelCount, "Nothing to do");
 130
 131   if (inputChannelCount > 6) {
 132     // Just drop the unknown channels.
 133     for (uint32_t o = 0; o < outputChannelCount; ++o) {
 134       ConvertAudioSamples(aInputChannels[o], aOutputChannels[o], aDuration);
 135     }
 136     return;
 137   }
 138
 139   // Ignore unknown channels, they're just dropped.
 140   inputChannelCount = std::min<uint32_t>(6, inputChannelCount);
 141
 142   const DownMixMatrix& m =
 143       gDownMixMatrices[gMixingMatrixIndexByChannels[outputChannelCount - 1] +
 144                        inputChannelCount - outputChannelCount - 1];
 145
 146   // This is slow, but general. We can define custom code for special
 147   // cases later.
 148   for (DstT* outChannel : aOutputChannels) {
 149     std::fill_n(outChannel, aDuration, static_cast<DstT>(0));
 150   }
 151   for (uint32_t c = 0; c < inputChannelCount; ++c) {
 152     uint32_t dstIndex = m.mInputDestination[c];
 153     if (dstIndex == IGNORE) {
 154       continue;
 155     }
 156     AddAudioSamplesWithScale(aInputChannels[c], aOutputChannels[dstIndex],
 157                              aDuration, m.mInputCoefficient[c]);
 158   }
 159   // Utilize the fact that in every layout, C is the only channel that may
 160   // contribute to more than one output channel.
 161   uint32_t dstIndex = m.mCExtraDestination;
 162   if (dstIndex != IGNORE) {
 163     AddAudioSamplesWithScale(aInputChannels[SURROUND_C],
 164                              aOutputChannels[dstIndex], aDuration,
 165                              m.mInputCoefficient[SURROUND_C]);
 166   }
 167 }
 168
 169 /**
 170  * UpMixMatrix represents a conversion matrix by exploiting the fact that
 171  * each output channel comes from at most one input channel.
 172  */
 173 struct UpMixMatrix {
 174   uint8_t mInputDestination[CUSTOM_CHANNEL_LAYOUTS];
 175 };
 176
 177 static const UpMixMatrix gUpMixMatrices[CUSTOM_CHANNEL_LAYOUTS *
 178                                         (CUSTOM_CHANNEL_LAYOUTS - 1) / 2] = {
 179     // Upmixes from mono
 180     {{0, 0}},
 181     {{0, IGNORE, IGNORE}},
 182     {{0, 0, IGNORE, IGNORE}},
 183     {{0, IGNORE, IGNORE, IGNORE, IGNORE}},
 184     {{IGNORE, IGNORE, 0, IGNORE, IGNORE, IGNORE}},
 185     // Upmixes from stereo
 186     {{0, 1, IGNORE}},
 187     {{0, 1, IGNORE, IGNORE}},
 188     {{0, 1, IGNORE, IGNORE, IGNORE}},
 189     {{0, 1, IGNORE, IGNORE, IGNORE, IGNORE}},
 190     // Upmixes from 3-channel
 191     {{0, 1, 2, IGNORE}},
 192     {{0, 1, 2, IGNORE, IGNORE}},
 193     {{0, 1, 2, IGNORE, IGNORE, IGNORE}},
 194     // Upmixes from quad
 195     {{0, 1, 2, 3, IGNORE}},
 196     {{0, 1, IGNORE, IGNORE, 2, 3}},
 197     // Upmixes from 5-channel
 198     {{0, 1, 2, 3, 4, IGNORE}}};
 199
 200 /**
 201  * Given an array of input channel data, and an output channel count,
 202  * replaces the array with an array of upmixed channels.
 203  * This shuffles the array and may set some channel buffers to aZeroChannel.
 204  * Don't call this with input count >= output count.
 205  * This may return *more* channels than requested. In that case, downmixing
 206  * is required to to get to aOutputChannelCount. (This is how we handle
 207  * odd cases like 3 -> 4 upmixing.)
 208  * If aChannelArray.Length() was the input to one of a series of
 209  * GetAudioChannelsSuperset calls resulting in aOutputChannelCount,
 210  * no downmixing will be required.
 211  */
 212 template <typename T>
 213 void AudioChannelsUpMix(nsTArray<const T*>* aChannelArray,
 214                         uint32_t aOutputChannelCount, const T* aZeroChannel) {
 215   uint32_t inputChannelCount = aChannelArray->Length();
 216   uint32_t outputChannelCount =
 217       GetAudioChannelsSuperset(aOutputChannelCount, inputChannelCount);
 218   NS_ASSERTION(outputChannelCount > inputChannelCount, "No up-mix needed");
 219   MOZ_ASSERT(inputChannelCount > 0, "Bad number of channels");
 220   MOZ_ASSERT(outputChannelCount > 0, "Bad number of channels");
 221
 222   aChannelArray->SetLength(outputChannelCount);
 223
 224   if (inputChannelCount < CUSTOM_CHANNEL_LAYOUTS &&
 225       outputChannelCount <= CUSTOM_CHANNEL_LAYOUTS) {
 226     const UpMixMatrix& m =
 227         gUpMixMatrices[gMixingMatrixIndexByChannels[inputChannelCount - 1] +
 228                        outputChannelCount - inputChannelCount - 1];
 229
 230     const T* outputChannels[CUSTOM_CHANNEL_LAYOUTS];
 231
 232     for (uint32_t i = 0; i < outputChannelCount; ++i) {
 233       uint8_t channelIndex = m.mInputDestination[i];
 234       if (channelIndex == IGNORE) {
 235         outputChannels[i] = aZeroChannel;
 236       } else {
 237         outputChannels[i] = aChannelArray->ElementAt(channelIndex);
 238       }
 239     }
 240     for (uint32_t i = 0; i < outputChannelCount; ++i) {
 241       aChannelArray->ElementAt(i) = outputChannels[i];
 242     }
 243     return;
 244   }
 245
 246   for (uint32_t i = inputChannelCount; i < outputChannelCount; ++i) {
 247     aChannelArray->ElementAt(i) = aZeroChannel;
 248   }
 249 }
 250
 251 }  // namespace mozilla
 252
 253 #endif /* MOZILLA_AUDIOCHANNELFORMAT_H_ */